Recent years have seen a dramatic increase in the number of hospitals providing coronary artery bypass surgery (CABG) services. This trend may provide improved access for patients but such access might be provided at a price. It is well known that high volume programs and high volume surgeons have, on average, better outcomes than programs and surgeons that perform fewer procedures. In addition, there is likely to be a 'learning curve' involved in any start-up program. There is little literature that directly addresses the question of the impact of a new cardiac surgery program on patients and their outcomes or oh the population at large. Medicare data are comprehensive, national in scope, provide information about operating institutions and physicians, and allow assessment of outcomes over time. We wish to use Medicare data to identify new cardiac programs over a ten year period, 1994-2003, and answer the following questions: 1) How do new cardiac programs address the 'learning curve' question? Do they select low risk patients, perform low risk procedures, recruit experienced surgeons? 2) How long does it take for new programs to move beyond the 'learning curve'? How quickly do new programs achieve acceptable minimum volume? Are outcomes different in new programs? How quickly do the outcomes in new programs approach those in established programs? 3) What happens to utilization of CABG in the population when a new program is initiated? Are new programs opened in markets where no other program exists? Does the addition of a new program to a local health care system result in a redistribution of existing procedures or result in an overall increase in the population-based rates of care, including CABG and total revascularization (either CABG or percutaneous coronary interventions) as additional capacity is brought online?